[backend] change cluster type to source files and read list from cfg db

Summary:
This avoids relying on the directories in infer-out/captured/ being created,
and instead gets the list of captured source files from the DB. This gives a
better type to clusters: `SourceFile.t` instead of `DB.source_dir`, which makes
the code a bit nicer too.

Reviewed By: jeremydubreil

Differential Revision: D6620460

fbshipit-source-id: c0edbf6

infer/src/IR/Cfg.ml

@@ -628,3 +628,10 @@ let pp_proc_signatures fmt cfg =
 let exists_for_source_file source =
   (* simplistic implementation that allocates the cfg as this is only used for reactive capture for now *)
   load source |> Option.is_some
+
+
+let get_captured_source_files () =
+  let db = ResultsDatabase.get_database () in
+  Sqlite3.prepare db "SELECT source_file FROM cfg"
+  |> SqliteUtils.sqlite_result_rev_list_step db ~log:"getting all source files"
+  |> List.filter_map ~f:(Option.map ~f:SourceFile.SQLite.deserialize)

infer/src/IR/Cfg.mli

@@ -64,3 +64,5 @@ val specialize_with_block_args :
 val pp_proc_signatures : Format.formatter -> t -> unit
 
 val exists_for_source_file : SourceFile.t -> bool
+
+val get_captured_source_files : unit -> SourceFile.t list

infer/src/backend/InferAnalyze.ml

@@ -77,23 +77,14 @@ let print_legend () =
 
 
 let cluster_should_be_analyzed ~changed_files cluster =
-  let fname = DB.source_dir_to_string cluster in
   (* whether [fname] is one of the [changed_files] *)
-  let is_changed_file =
-    (* set of source dirs to analyze inside infer-out/captured/ *)
-    let source_dirs_to_analyze changed_files =
-      SourceFile.Set.fold
-        (fun source_file source_dir_set ->
-          let source_dir = DB.source_dir_from_source_file source_file in
-          String.Set.add source_dir_set (DB.source_dir_to_string source_dir) )
-        changed_files String.Set.empty
-    in
-    Option.map ~f:source_dirs_to_analyze changed_files
-    |> fun dirs_opt -> Option.map dirs_opt ~f:(fun dirs -> String.Set.mem dirs fname)
-  in
+  let is_changed_file = Option.map changed_files ~f:(SourceFile.Set.mem cluster) in
   let check_modified () =
-    let modified = DB.file_was_updated_after_start (DB.filename_from_string fname) in
-    if modified && Config.developer_mode then L.debug Analysis Medium "Modified: %s@." fname ;
+    let modified =
+      DB.source_dir_from_source_file cluster |> DB.source_dir_to_string |> DB.filename_from_string
+      |> DB.file_was_updated_after_start
+    in
+    if modified then L.debug Analysis Medium "Modified: %a@\n" SourceFile.pp cluster ;
     modified
   in
   match is_changed_file with
@@ -128,7 +119,7 @@ let main ~changed_files ~makefile =
       (* delete all specs when doing a full analysis so that we do not report on procedures that do
          not exist anymore *)
       if not Config.reactive_mode then DB.Results_dir.clean_specs_dir () ;
-      let all_clusters = DB.find_source_dirs () in
+      let all_clusters = Cfg.get_captured_source_files () in
       let clusters_to_analyze =
         List.filter ~f:(cluster_should_be_analyzed ~changed_files) all_clusters
       in
@@ -142,7 +133,7 @@ let main ~changed_files ~makefile =
       let is_java =
         lazy
           (List.exists
-             ~f:(fun cl -> DB.string_crc_has_extension ~ext:"java" (DB.source_dir_to_string cl))
+             ~f:(fun cl -> Filename.check_suffix ".java" (SourceFile.to_string cl))
              all_clusters)
       in
       print_legend () ;

infer/src/backend/cluster.ml

@@ -14,7 +14,7 @@ module F = Format
 (** Module to process clusters of procedures. *)
 
 (** a cluster is a file *)
-type t = DB.source_dir
+type t = SourceFile.t
 
 (** type stored in .cluster file: (n,cl) indicates cl is cluster n *)
 type serializer_t = int * t

infer/src/backend/cluster.mli

@@ -13,7 +13,7 @@ module F = Format
 (** Module to process clusters of procedures. *)
 
 (** a cluster is a file *)
-type t = DB.source_dir
+type t = SourceFile.t
 
 (** type stored in .cluster file: (n,cl) indicates cl is cluster n *)
 type serializer_t = int * t

infer/src/backend/clusterMakefile.ml

@@ -42,7 +42,7 @@ let create_cluster_makefile (clusters: Cluster.t list) (fname: string) =
   let outc = Out_channel.create fname in
   let fmt = Format.formatter_of_out_channel outc in
   let do_cluster cluster_nr cluster =
-    F.fprintf fmt "#%s@\n" (DB.source_dir_to_string cluster) ;
+    F.fprintf fmt "#%a@\n" SourceFile.pp cluster ;
     Cluster.pp_cluster fmt (cluster_nr + 1, cluster)
   in
   pp_prolog fmt clusters ;

infer/src/backend/exe_env.ml

@@ -202,7 +202,8 @@ let get_proc_desc exe_env pname =
 (** Create an exe_env from a source dir *)
 let from_cluster cluster =
   let exe_env = create () in
-  add_cg exe_env cluster ; exe_env
+  add_cg exe_env (DB.source_dir_from_source_file cluster) ;
+  exe_env
 
 
 (** [iter_files f exe_env] applies [f] to the filename and tenv and cfg for each file in [exe_env] *)